Evolution is the ultimate design algorithm behind biology and if sped up, could have enormous utility in bioengineering. I propose a unique strategy, called orthogonal replication (OrthoRep), to achieve fast and scalable targeted gene evolution in vivo so that the evolutionary process can be routinely applied for biomolecular engineering. The idea is to create a cell, starting with yeast, that has a second replication system consisting of a special DNA plasmid replicated by a dedicated DNA polymerase (DNAP). The second system would be orthogonal to genomic replication such that the dedicated DNAP (ortho-DNAP) only replicates the special plasmid (ortho-plasmid) and not the host genome. The ortho-DNAP could then be engineered to mutate the ortho-plasmid at rates far exceeding what the genome could tolerate. Our analysis suggests that OrthoRep could accelerate evolution by enormous amounts, as a gene encoded on the ortho-plasmid could in principle be forced to diversify ~106-fold faster than if it were encoded on the genome of yeast. We have successfully established OrthoRep in Saccharomyces cerevisiae, and we will continue its development by engineering highly error- prone ortho-DNAPs to reach maximum rates of asexual gene evolution. We will also add sexual evolution to ortho-plasmid, spurred by fortuitous observations of our ortho-plasmids' natural tendency to recombine. Finally, we will engineer copy number control for the ortho-plasmid to enable a broader array of selectable functions, especially negative selections. Once OrthoRep is developed, I propose to apply it to the rapid evolution of in vivo biosensors. Metabolic engineering has great potential for biomedicine, as it promises to move multi-gene biosynthetic pathways that synthesize complex drugs from dif?cult natural sources into cheap microbial production hosts. However, when a multi-gene biosynthesis pathway is transferred into a microbe, an array of rational and combinatorial optimization steps are inevitably needed. Optimization requires the ability to detect product production, but there are no high-throughput assays capable of this. Our OrthoRep system will remove this key roadblock in metabolic engineering by evolving in vivo biosensors for small molecules of interest nearly on-demand. We will evolve in vivo biosensors for four molecules, taxadiene, casbene, amorphadiene, and parthenolide, whose ef?cient production in yeast offer different challenges to metabolic engineering. Not only will these biosensors be directly useful for microbial production of these drug and drug precursors, lessons learned will help us solidify OrthoRep as a potentially transformative evolutionary engineering technology that can be broadly applied.